2.1. Study design and setting

An open-label, cluster-randomized trial was conducted in the tribal block of Jawadhu Hills (JH) in southern India from September 2013 to December 2016. The JH block borders the Vellore and Tiruvannamalai districts of Tamil Nadu and is located on hilly terrain with poor road access, lack of safe drinking water, and poor sanitation facilities [16]. It is predominantly occupied by tribal agrarian communities, many of whom temporarily migrate to coffee plantations in the neighboring states as short-term laborers during non-agricultural seasons. The predominant STH species is hookworm, with the prevalence and intensity of infection increasing with age [10]. From 2007 to 2015, residents aged ≥1 year typically received multiple rounds of annual MDA with diethylcarbamazine (DEC) and albendazole as part of a national program for the elimination of lymphatic filariasis [17].

2.2. MDA strategies

The trial design, data collection methods, and follow-up have been presented in detail elsewhere [16]. Briefly, a cluster-randomized trial was conducted in 45 tribal villages, which were randomized into three groups (1:1:1): (a) annual community-wide MDA (one-cycle group); (b) two cycles of community-wide MDA at one-month intervals (two-cycle group); and (c) four cycles of community-wide MDA – two cycles one month apart at the beginning, followed by another two after six months (four-cycle group). The treatment intervals in the two- and four-cycle groups were based on hookworm biology: the two-cycle group covered the larvae’s extrinsic incubation period in the soil to reduce reinfection [15], whereas the four-cycle group was added to account for the survival of infective larvae for several weeks in the soil under favorable environmental conditions [18].

Eligible participants received anthelminthic treatment with albendazole (400 mg) distributed by trained field workers through door-to-door household visits. Individuals with a history of seizure, epilepsy, known neurological disorder, known history of hypersensitivity to albendazole, serological evidence of HIV infection, immune-compromise or currently pregnant were excluded due to concerns around the perceived risks of treatment in these groups. Participants were interviewed the subsequent day to estimate treatment coverage and were followed up for three consecutive days and again after one-week post-treatment to record any adverse/serious adverse events following the intervention. Households that were not available for treatment were revisited the next day. Households were excluded if the dwelling could not be found, was vacant, or had no adult resident at home on three consecutive visits. The participants and trial personnel were not masked to allocation due to the nature of the trial, although the laboratory personnel analyzing the stool samples were unaware of the treatment allocation status.

The treatment (MDA) strategy was at the community-level; however, the study objectives pertain to the individual participant level. The unit of randomization was a tribal village with 15–100 households. Villages proximal to motorable roads were preferred, anticipating adverse neurological events associated with asymptomatic neurocysticercosis in pork-consuming communities, such as the trial population [19,20]. The study’s enrollment started in October 2013 and was completed across all 45 villages by November 2014. The MDA (per protocol) was completed in November 2014 and follow-up data collection for all study villages was completed by November 2016. The ‘National Deworming Day’ program (bi-annual targeted deworming of PSAC and SAC children between 1 and 19 years of age, held every February and August), which commenced from February 2015 [21], continued concurrently in all study villages towards the end of the follow-up period.

2.3. Baseline census

The baseline demographic and socio-economic characteristics of the residents were captured through a door-to-door household census of the study villages [16]. Trained interviewers administered structured questionnaires to all participants (primary caregivers in case of minors) to obtain information on demographic and employment history (including details of agriculture-related activities), footwear usage, defecation and handwashing practices, type of flooring, and animal contact. The respondents’ socio-economic status (SES) was calculated using a locally developed and validated SES scale that accounts for caste, type of housing, land ownership, educational and occupational status of the head of the household [22]. Based on the SES score tertile, respondents were classified as belonging to low, medium, or high SES households.

2.4. Coprological surveys

Cross-sectional coprological surveys were conducted at baseline, after every MDA cycle, and every three months post-MDA, for one year, and at the end of two years to ascertain the prevalence and intensity of hookworm infection. The follow-up surveys were standardized relative to the timing of the final MDA round within each intervention arm, beginning three months after the single MDA round in one-cycle villages, three months after the second round in two-cycle villages (i.e., four months after baseline), and three months after the fourth round in four-cycle villages (i.e., eight months after baseline) (see Fig 2 of [16]).

Participants for the coprological surveys were randomly selected from among the permanent residents of the study villages aged 2–70 years (both years inclusive) who met the eligibility criteria; exclusions included syndromic or serological evidence of HIV infection or other immunocompromising conditions, pregnancy, a history of seizures, epilepsy or other neurological disorders, and known hypersensitivity to albendazole. The door-to-door baseline census provided the sampling frame. For each survey round, eligible residents were randomly ordered, with the first 30 participants from each village, who provided written informed consent and at least one stool sample, enrolled.

Selected participants provided up to three stool samples, collected in a plastic container labeled with a unique identification number. Individuals who could not provide a sample immediately were contacted again later that day or the next day. Stool samples were stored in a fridge in the field office (4–8 degrees Celsius) and transported to Christian Medical College, Vellore, periodically (2–3 times a week within 72 hours of collection), based on the number of samples collected. They were then examined microscopically for the presence of hookworm ova using saline and wet iodine preparation [23]. A senior technician re-examined 10% of slides for quality control. Samples positive for hookworm ova were then tested by the McMaster method to measure the number of eggs per gram of stool (EPG) to estimate the intensity of infection [24].

2.5. Outcomes

The primary outcome measure was the effectiveness of treatment, evaluated based on the prevalence and intensity of infection at the end of 12 months after the last MDA cycle among all the enrolled individuals. Secondary outcome measures included prevalence and intensity of infection after each treatment cycle at 3-, 6-, 9- and 24-month post-MDA.

2.6. Statistical analysis

Data were analyzed using STATA version 14.2 for Windows (StataCorp, College Station, Texas) and R version 3.2.2 (http://www.r-project.org/). All variables were examined using descriptive statistics (measures of central tendency [means and medians], dispersion [standard deviations, interquartile ranges]) for continuous variables, frequency counts, and marginal percentages [with 95% confidence intervals] for categorical variables). The success of randomization was evaluated by comparing the baseline characteristics of the groups without formal statistical testing.

The prevalence and mean intensity of infection were calculated for each village. The age-prevalence and age-intensity profiles of hookworm infection were explored by computing age-specific estimates of hookworm prevalence and intensity. A non-parametric Wilcoxon-type test for trend was used to determine the change in prevalence and intensity of infection with age [25]. Taylor linearized standard errors accounted for the clustered data structure [26]. The baseline demographic and behavioral factors associated with hookworm infection were identified using logistic regression analysis and odds ratios (OR) with 95% confidence intervals (CI) calculated; cluster-robust standard errors accounted for the non-independence of observations [27].

The primary trial analysis was an intention-to-treat analysis comparing the prevalence and intensity of hookworm infection between the three groups at the end of 12 months after the last MDA cycle. Generalized estimating equations (GEE) with a binomial distribution and logit link function for hookworm prevalence and negative binomial distribution with log link function for hookworm intensity were fitted. Exchangeable correlation structure and robust standard errors were used, and age, gender and baseline hookworm prevalence/mean intensity were adjusted for in all models. The effect of clustering at the village and the household level was explored through multilevel (hierarchical) modeling.

3.1. Enumeration of the study villages, enrolment, follow-up, and MDA coverage

The baseline survey identified a total population of 11857 residents in the 45 study villages, with a median (IQR) of 238 (203–289) population per village [16]. The spatial distribution of the study villages, by intervention group, is presented as S1 Fig.

On comparing the socio-demographic profile of the households in the study villages, collected through the baseline survey, there were more households from the low SES category residing in the single-cycle group (30.6%) compared to the two- (26.6%) and four-cycle groups (25.5%). The proportion of households with at least one member engaged in agriculture-related activities (94.2%), or with a public tap or borewell as the primary source of drinking water (89.4%) in the two-cycle villages was higher compared to the other two groups (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of baseline characteristics of the study villages assigned to different MDA regimens.

https://doi.org/10.1371/journal.pntd.0013440.t001

Of the 11857 residents, 8681 (73.2%) fulfilled the eligibility criteria and consented to participate in the trial. The primary reason for non-participation was migration outside the study area for temporary employment: 1947/11857 (16.4%) residents were absent in their respective villages during the study period, as illustrated in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Overview of participant enrollment and implementation of the mass drug administration (MDA) regimens.

https://doi.org/10.1371/journal.pntd.0013440.g001

The per-protocol MDA coverage (all doses of albendazole received as per the allocated intervention) was highest in the single-cycle group (85.6%, 2458/2872), followed by two- (77%, 2010/2612) and four-cycle (50.4%, 1612/3197) groups. However, the overall MDA coverage (at least one dose received) was slightly higher in the two-cycle group (94.8%, 2475/2612) compared to the four-cycle (93.8%, 2998/3197) group (Fig 1). The individuals who missed MDAs were more likely to be teenagers (15–19 years) or adult males (20–54 years), reflecting the migratory pattern of the study population. These demographic subgroups (teenagers and adult males) were also less likely to participate in the coprological surveys (S2 Fig).

3.2. Baseline epidemiology of hookworm infection

A total of 2082 participants (median [IQR] of 46 [44–48] participants per village) provided one or more stool samples; 1727 (83%) provided all three stool samples as per protocol. There was a wide variation in the prevalence of hookworm infection between the villages, ranging from 2.1- 44.2%. The distribution of the intensity profile of hookworm infection showed a highly aggregated behavior: only 1.2% of the sampled population had moderate-to-high intensity infection (MHI, ≥ 2000 EPG). Both the prevalence and intensity of hookworm infection increased with age (S1 Table). Also, males (20.3%, 95% CI: 17.3-23.6%) had a higher prevalence of hookworm infection than females (16.9%, 95% CI: 13.9-20.3%), although the intensity of infection was comparable between both sexes (Mean [SE] EPG = 125 [26] and 126 [28] for males and females, respectively). Analysis of the intensity of infection in hookworm-infected participants revealed that those with all three positive samples had the highest egg counts (mean EPG: 3777, SE: 735), followed by those with two positive samples (mean EPG: 896, SE: 217) and one positive sample (mean EPG: 192, SE: 21), respectively (Kruskal Wallis test, P < 0.001). The baseline (pre-intervention) prevalence of hookworm infection was 14.8% in the single-cycle group, 18.5% in the two-cycle group and 22.4% in the four-cycle group (Table 2). The mean (SE) EPGs were 52 (10), 144 (63) and 181 (39) in single, two-, and four-cycle groups, respectively.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Pre- and post-intervention prevalence and intensity of hookworm infection in the study villages assigned to the different MDA regimens.

https://doi.org/10.1371/journal.pntd.0013440.t002

In the multivariable analysis, male gender (aOR: 1.31, P = 0.032) and living in a house with mud flooring (aOR: 1.31, P = 0.038) were associated with an increased risk of hookworm infection; as was older age, with the highest risk in participants aged 45–54 years (adjusted odds ratio [aOR]: 5.26, P < 0.001). Drinking treated water was protective (aOR: 0.75, P = 0.049). Regular contact with animals was significantly associated with hookworm infection in the univariate (OR: 1.40, P = 0.030) but not in the multivariable analysis (aOR: 1.06, P = 0.726). Footwear usage, washing hands with soap and water after defecation, performing agriculture-related activities in the preceding six months, and socioeconomic status were not associated with hookworm infection risk (S2 Table).

3.3. Post-MDA prevalence and intensity of hookworm infection

After the first MDA cycle, the prevalence of hookworm infection decreased to 3.7% (95% CI, 2.6-5.1%), i.e., a relative reduction of 80% (72.4-85.9%) from the baseline prevalence of 18.5% [16], further decreasing to 1.8% (95% CI, 1.0-3.3%) after the second MDA cycle. Thereafter, the prevalence remained stable despite two additional MDA cycles after six months (1.3% [95% CI: 0.6-2.8%] and 1.6% [0.6-4.3%] after the third and fourth MDA cycles, respectively).

During the post-intervention period (3–24 months), the prevalence of hookworm infection ranged from 2.6%-5.2% in the single-cycle group and from 7.9-4.0% in the two-cycle group. On the other hand, the prevalence of hookworm infection in the four-cycle group was constantly lower, ranging from 1.0-2.6% (Table 2). The post-intervention mean EPG of feces followed a similar pattern, ranging from 4-14 EPG in the single-cycle group, 13–18 in the two-cycle group and 2–8 in the four-cycle group (Table 2).

3.4. Assessment of MDA strategies

The post-MDA GEE analysis (adjusted for baseline prevalence/intensity of hookworm infection, age, and gender) showed no significant difference in the prevalence and intensity of hookworm infections between the single- and the two-cycle groups at 3-, 6-, 9-, 12- and 24-month post-intervention (Table 3). However, the prevalence of hookworm infection in the four-cycle group was significantly lower at 3-month (OR: 0.62, P = 0.024), 6-month (OR: 0.41, P = 0.001) and 12-month (OR: P = 0.016) post-intervention when compared to the single cycle group. Moreover, in the four-cycle group, the intensity of infection was significantly lower (P < 0.05), as compared to the single-cycle group, at 3-, 6-, 9-, and 12-month post-intervention. There was, however, no statistically significant difference in the prevalence and intensity of infection between the three groups (one, two and four cycle) 24-month post-intervention (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Results of GEE analysis to ascertain the effect of the intervention (MDA) on hookworm prevalence and intensity.

https://doi.org/10.1371/journal.pntd.0013440.t003

3.6. Adverse and serious adverse events

Overall, 21 adverse events (AE) were reported in 18 participants (three had two AEs each) during the study period. The commonly reported mild and moderate symptoms were gastritis (14/21), fever with rash or myalgia (2/21), urticaria (1/21), and giddiness (1/21). Three severe adverse events (SAE) were attributed to acute appendicitis, basal cell carcinoma of the left eye, and post-streptococcal glomerulonephritis with nephrotic syndrome, of which the latter succumbed to the illness. None of the SAEs were associated with the intervention.